Research On Building Settlement Monitoring Based On Combined Model

With the development of the 13th five year plan to the 14th five year plan,the speed of urban development is more and more rapid.Due to various reasons,accidents caused by building settlement often occur around us,which not only cause inconvenience to people’s lives,but also bury hidden dangers to people’s life safety.The rapid development of building settlement monitoring research has become particularly important.Using advanced surveying and mapping instruments,the settlement of buildings is monitored regularly according to the standards and specifications,certain mathematical processing methods are used to scientifically process the obtained data,so as to find a mathematical model that can scientifically reflect the law of building settlement and correctly predict the building settlement.Taking Hefei Shizhang primary school as an example,use Matlab programming language is used to design GUI program interface.Combined with the actual engineering data,the gray theory model,time series model and artificial neural network model are adopted to predict the settlement.By analyzing and comparing the residual and RMSE between the measured value and the predicted value,the advantages and disadvantages of the three models in this engineering example are judged.The first mock exam of three models in this project is analyzed,and the idea of combination model is put forward.Finally,the combination of grey time series model and grey neural network model is used to find the most suitable prediction model for building settlement monitoring.The comparison results show that in the grey time series combination model,no matter which method is used to predict the later data,the error will become larger and unstable with the extension of time,because the parameters are fixed with the passage of time,so it is difficult to meet the later prediction.However,in the grey neural network combination model,with the extension of time,the error becomes smaller and relatively stable.In addition,the predicted value is also very close to the measured value,and it is better than the artificial neural network model with the best effect in the single prediction model.This paper discusses the different characteristics of different combination models in historical data fitting and unknown data prediction,summarizes the general rules and applicability of various combination models,and puts forward better prediction models.At the same time the shortcomings of the first mock exam model are improved,and the optimal combination model is combined to get more accurate prediction results.It can provide necessary information for the safety diagnosis of surface subsidence,so as to find out the problems in time and take measures.Figure[19]Table[16]Reference[60]...